OPERATIONAL WAVE FORECASTING FOR TROPICAL CYCLONES CONDITIONS

Transcription

1 OPERATIONAL WAVE FORECASTING FOR TROPICAL CYCLONES CONDITIONS Jean-Michel Lefèvre 1, Lotfi Aouf 1, Pierre Queffeulou 2, Abderrahim Bentamy 2, Yves Quilfen 2 1 Meteo-France, Toulouse, France 2 IFREMER, Plouzané, France with support of CNES, and ANR funding (HEXECO project)

2 Motivation Meteo-France s Responsibilities International: GMDSS and RMSC National: Warning System recently extended New Wave Forecasting System: MFWAM Global Validation through JCOMM/WFVS Regional Validation using data from JCOMM/WFVS Tropical Cyclone Validation- Errors estimation Benefit for applications such as High sea states prediction Unexpected waves (Spectral shape) Wave Setup warning (beach slope, HsxT 2 )

3 Challenge: errors due to unknown growth curves, Snl approximation (use of DIA, Hendrik talk), underestimated maximum wind speed- TC location.uncertainty Groen and Dorrenstein 76

4 Methodology Implementation of MFWAM model (ECWAM code with new physics) over part of Indian Ocean nested in Global MFWAM model driven by a Limited Area NWP model with TC bogusing (ALADIN) Simulation over 3 TC seasons ( ) using MF operational wind forcing (ALADIN) but also with ECMWF wind analyses, ALADIN wind analyses and Blended Model-SCAT Winds. Collocation of model data with Jason-1,Jason-2, ERS-2, ENVISAT and GFO altimeters data : inter-calibration+ averaging (noise reductionrepresentativness error reduction by box averaging 1 ) Computation of annual mean biases, NRMSE for SWH, analyse of some individual cases

5 Summary/Conclusions MFWAM-437 underestimates wave heights for TC conditions when driven with LAM ALADIN and ECMWF winds. MFWAM-441 produces unbiased SWH for high winds conditions, including TC conditions, when driven by ALADIN, ECMWF and Blended winds (SCATT+model first guess) All wind forcings produce relatively accurate sea-state estimates (typically 10% random error up to 7m SWH) with very small biases (< 5%) Improvement when using Blended winds Closer to scatt winds than model analyses: more spatial variability in blended winds

6 International duties:france within the GMDSS Global Marine Distress and Safety System RMSC La REUNION Issuing Service for Metarea II and Metarea VIII(S) II III Preparation Service for Metarea III (W), Metarea VII VII VIIIS MF LA REUNION: Regional Specialized Met Center (RMSC): Tropical Cyclone Monitoring

7 Extended warning system for large waves and coastal inundation For the public, warnings are Information on the level of potential danger in an area, if orange and red, this means that people are invited to : - read associated bulletins to learn more about - on going events and their current and future evolution - possible consequences - recommendations from authorities about what to do - stay informed of the messages from authorities

8 Example of Warning Map High waves and coastal flooding warning system operational since october 3 rd this year. Decision to speed up the setting taken after Xynthia storm which was associacted with dramatic coastal flooding

9 Example of event that activated the warning system in French Polynesia: 5 m and 18s (wave length 500 m) swell Expected wave set up, locally more than 1.5 m No buoy data, only data from space

10 New wave forecasting system of Meteo- France: MFWAM, thanks to joint efforts with : ECMWF (source code-validation with buoy data) SHOM (Physics) IFREMER (Validation with Altimeter data) CNES (Data assimilation-validation) -MFWAM based on ECWAM source code modified for new wave physics- mainly the dissipation term (Ardhuin et al. 2010) -Introduction of ASAR LP2 and Altimeter data (Aouf et al. 2010) -Implementation of Multi-grid nesting: from Global to Regional models -Introduction of a partitioning scheme for swell components

11 New Dissipation terms from Ardhuin et al. (2010) Non isotropic dissipation : -> Better adjustment of the mean direction and angular spreading -> Breaking: threshold mechanism from the saturation spectrum 0.7 f C c3f( f, ).max r df C ' 0 0 B( f ', ') B,0. d ' ' g Last term for the dumping of short waves by dominant waves (term cumulative term, C3=0.4 in 441 C3=1 in 437)

12 Modification of the input source term (Ardhuin et al.2010) Drag limitation stress reduction for MFWAM-441 to adjust with new dissipation based on saturation spectrum Su=1 for MFWAM-441 Su=0 for MFWAM-437

13 New wave forecasting system of Meteo-France: 2 global wave models MFWAM driven by different wind forcings: ECMWF and ARPEGE/IFS - resolution 55km 1 regional wave model MFWAM resolution 10km driven by ARPEGE winds soon with ECMWFwinds aswell MFWAM-ALADIN-REUNION resolution 0.25 Wind forcing ALADIN-REUNION grille 0.1 km- 3hourly winds with TC bogusing

14 ALADIN-Réunion Model Operational since nov km horizontal résolution Observation dataset SYNOP/SHIP/BUOY, Radiosondes, Profilers, Aircrafts Cloud winds (METEOSAT 7, 9) Scatterometer winds (QuikScat, ERS-2 and METOP-A) ATOVS HIRS, AMSU-A, AMSU-B (NOAA 15,16,17,18 ; AQUA ; METOP-A), AIRS (AQUA) SSMI in clear-sky conditions (DMSP F13, F14) Satellite GPS But Lack of observation in the Hurricane 3D Var assimilation scheme o a 6h window (4/day) 60 vertical levels 2-day leadtime forecast (at 0 and 12h UTC). 3-day very soo BC from Arpege model. pseudo- 200 km 100 km

15 Systematic Verification of the Wave Forecasting System: 1. ECMWF/WFVS (Wave Forecasting Verification System) based on Marine Automatic Weather Stations (MAWS) Problem, No MAWS in our area of interest 2. ESA/GlobWave Pilot System to extend the WFVS to altimeter data Following WMO/JCOMM recommandation

16 Monthly WFVS reports from JCOMM web site Forecast range in days MF----- ECMWF NCEP _ September 2011 Relative error in %

17 Forecast range in days MF----- ECMWF NCEP _ September 2011 Relative error in %

18 Hurricane Katia, september 2011

19

20 Regional Europe MFWAM 11 km resolution

21 Modelling Waves from TC requires appropriate wind forcing and Wave model: - realistic winds at typical scale of km - realistic wave growth at high winds MFWAM- ALADIN

22 Validation over seasons 2008 and 2009 done, 2010 ongoing End 2008-Mid 2009 saison

23 Altimeter data Model improvment ALTIMETER DATA BASE: IFREMER ERS-2 ENVISAT JASON1-2 GFO Intercalibrated data averaged along track over 1 in latitude First Version of MFWAM (437, see Ardhuin et al. 2010,JPO): overestimation of high waves with all wind forcings 19 January 2012 No. 23

24 Implementation of MFWAM-441 (see Ardhuin et al. 2010, JPO) - Drag Limitation (Zo max) - Shelter term added in the input source term - Dissipation rate modified - Tested with 3 sources of wind:. Aladin. ECMWF. Blended (Scatt + ECMWF guess)

25 ECMWF ALADIN Biais reduction (blue curve) BLENDED

26 ECMWF Case with small differences however ALADIN Blended better BLENDED

27 ECMWF Large bias ALADIN Small bias but shifted position ---> increases rmse BLENDED No bias right position

28 Concluding Remarks/Perspectives Importance of satellite data for validation, calibration of model parameterizations > ESA/GlobWave Pilot System to extend the WFVS to altimeter data Following WMO/JCOMM recommandation Need to extent the comparisons to other periods (2010 and 2011) and other areas (Northern Atlantic) in order to study more cases with larger wave heights and Compare NWP winds with winds derived from altimetry (IFREMER wind Algorithm, see F. Ardhuin s presentation) Use of SAR data to validate peak periods and directions (and Significant Wave Heights CWAVE algo. From DLR, see T. Bruns s presentation) Regional models for the Caribbean, French Polynesia and New Caledonia areas will be implemented in For most of those areas no in situ data are available. Use of a 2.5 km resolution NH NWP model (AROME) currently tested over La Reunion Area